Diversity Dependent Uncertainty Management for Hydrocarbon Stimulation in Uncertain Heterogeneous Reservoir with Improved Efficiency

Abstract: The history of oil and gas well stimulation through hydraulic fracturing is characterized by a pursuit of optimal designs tailored to reservoir properties. However, as with many engineering systems, the impact of variability and uncertainty (in this case of reservoir properties) is not accounted in these deterministic approaches. We propose an expansion of the principle of design diversity wherein the focus becomes developing optimal portfolio combinations of multiple designs rather than repeated application of a single design. This approach can substantially increase efficiency and decrease risk when designing systems where outcome desirability is definable as a smoothly-varying efficiency and where risk is related to the variance of outcomes tied to uncertain variability of the design environment. For the case of oil/gas wells, we demonstrate diverse portfolios of designs can improve stimulation efficiency by nearly a factor of four and cut risk associated with variability and uncertainty of outcomes by over 80%. Hydraulic fracturing optimization is therefore fundamentally dependent upon uncertainty, and a change of approach toward the pursuit of design diversity can have a profound impact on efficiency and risk, and in turn cost and environmental impacts, in oil and gas development.

• The paper introduces a portfolio-based approach inspired by Modern Portfolio Theory to incorporate uncertainty in hydraulic fracturing design.
• It demonstrates up to a 3.8-fold increase in energy efficiency and over 80% risk reduction compared to traditional deterministic designs.
• The study emphasizes leveraging design diversity to optimize well performance under heterogeneous reservoir conditions.

Diversity-Driven Uncertainty Management for Hydraulic Fracturing

Introduction and Motivation

Hydraulic fracturing (HF) in heterogeneous subsurface reservoirs is dominated by large uncertainties primarily due to spatial variability in reservoir properties and limitations in subsurface characterization. Conventional deterministic optimization methods, including practices such as extreme limited entry (EXL), target specific designs but often disregard or incompletely address stochasticity in reservoir behavior. This can lead to inefficiencies, increased economic and environmental costs, and suboptimal resource extraction. This paper proposes a paradigm shift towards explicit uncertainty management by leveraging design diversity through portfolio optimization. The conceptual framework borrows from Modern Portfolio Theory (MPT) and Capital Market Link (CML) in finance, embedding the risk–return tradeoff directly into the engineering design process.

Methods: Portfolio Theory Adaptation

The HF design space is explored by generating large sets (~7200 per category) of candidate treatments with uniform and non-uniform cluster spacings, accompanied by stochastic realizations of critical rock and operational parameters. The essential innovation is the definition of each design's “return” (energy efficiency—fracture surface area per input energy) and “risk” (composite measure of intra-well and inter-well variance in performance due to heterogeneity).

The MPT framework is adapted such that:

• Each unique HF design substitutes for a financial asset.
• Portfolios correspond to sets of designs applied across a development, with weights determining the proportion of wells treated by each design.
• The covariance structure is nearly diagonal, as each well is treated independently and rock property variations across wells are uncorrelated at tractable scales.

The expected portfolio return and risk are computed by linearly combining design returns and risks. The EXL methods, while low-yield but low-variance, provide analogs to risk-free assets in CML, serving as anchors for constructing efficient frontiers via tangent line analysis through the Pareto set derived from high-efficiency and low-risk portfolios.

Numerical Results and Key Findings

Quantitative simulations demonstrate that constructing portfolios with diverse HF designs outperforms repeated single-design application in both efficiency and risk mitigation. The major, explicit findings include:

• Efficiency gains: Portfolios of high-efficiency designs yield up to 3.8-fold increases in energy efficiency over single repeated designs.
• Risk reduction: Portfolio combinations using EXL can achieve over 80% reduction in risk relative to baseline deterministic designs.
• Non-uniform fracture cluster spacing provides marked benefits in risk reduction at low to moderate heterogeneity (5–10%), but the advantage diminishes as reservoir variability increases (e.g., at 20%, spatial randomness dominates).
• High variability in perforation pressure loss and injection rates within the portfolios, previously considered insensitive in deterministic frameworks, are shown to be critical levers for risk reduction under uncertainty.

These findings remain robust over ranges of rock property uncertainty and demonstrate that diversity-based design radically alters the optimal control landscape compared to deterministic or even stochastic optimization applied with single-design focus.

Theoretical and Practical Implications

This work challenges the classical doctrine of field “factory mode” with a single optimal design, which is shown to be suboptimal in the presence of inevitable heterogeneity. By analogizing engineering design to portfolio allocation, it formally incorporates the effect of uncertainty and heterogeneity in operational decision making. Key implications are:

• Efficient HF strategies in highly heterogeneous reservoirs cannot be single-point deterministic solutions; optimal development requires blending multiple design archetypes.
• Substantial improvements in cost, energy intensity, and environmental footprint can be obtained not through more data acquisition or characterization, but purely by embracing and optimizing design diversity.
• The generality of the portfolio approach applies to any engineering domain combining continuous performance metrics and strong spatial uncertainty: e.g., wind farm siting, agricultural treatment deployment, and large-scale vaccine rollout.

Outlook and Future Directions

This diversity-based portfolio design approach is extensible and amenable to integration with optimization metaheuristics, machine learning-based surrogate models for fast evaluation, and real-time adaptive management frameworks. Future developments may consider:

• Incorporation of secondary outcomes such as induced seismicity, water use, and CO$_2$ footprint as vector-valued returns.
• Dynamic portfolio rebalancing as new information accrues during field development (“active learning”).
• Method transfer to renewable energy, agriculture, and epidemiology where uncertain, heterogeneous environments challenge conventional deterministic or scenario-based optimization.

Conclusion

A portfolio optimization framework informed by MPT/CML, when applied to HF design, fundamentally redefines uncertainty management in subsurface engineering. Substantial gains in efficiency and reductions in risk are achievable even with minimal additional data or reservoir characterization. The results underscore the inadequacy of repetitive deterministic design application under uncertainty, instead advocating for design mixture strategies that optimize both mean outcomes and tail risk. This approach possesses broad implications well beyond hydrocarbon production, providing an extensible method for rational engineering under heterogeneity and uncertainty.

---

Reference: "Diversity Dependent Uncertainty Management for Hydrocarbon Stimulation in Uncertain Heterogeneous Reservoir with Improved Efficiency" (2011.13520)